Power Output Monitoring and Anomaly Identification of Photovoltaic Systems Using Graphical Modeling

Photovoltaic cells are being increasingly installed worldwide as a type of important renewable resource for power generation. Among the installed photovoltaic systems, data are collected from the control and monitoring system, and the anomaly detection researches are mostly based on the individual photovoltaic cells and the systematic behavior in a group of photovoltaic systems is not well studied. Based on the assumption that the photovoltaic cells installed in the same location are with very close solar radiation, the power output of each subsystem is considered in this work for monitoring the operating status of the subsystem and identifying the anomalous subsystem, where spatiotemporal pattern network, a probabilistic graphical modeling approach, is applied. The results show that the presented framework is able to detect the anomaly and locate the anomalous subsystem.